Cognitive remediation (CR) benefits cognition and functioning in psychosis but we do not know the optimal level of therapist contact, so we evaluated the potential benefits of different CR modes. A multi-arm, multi-center, single-blinded, adaptive trial of therapist-supported CR. Participants from 11 NHS early intervention psychosis services were independently randomized to Independent, Group, One-to-One, or Treatment-as-usual (TAU). The primary outcome was functional recovery (Goal Attainment Scale [GAS]) at 15-weeks post randomization. Independent and TAU arms were closed after an interim analysis, and three informative contrasts tested (Group vs One-to-One, Independent vs TAU, Group + One-to-One vs TAU). Health economic analyses considered the cost per Quality Adjusted Life Year (QALY). All analyses used intention-to-treat principles. We analyzed 377 participants (65 Independent, 134 Group, 112 One-to-One, 66 TAU). GAS did not differ for Group vs One-to-One: Cohen's d: 0.07, -0.25 to 0.40 95% CI, P = .655; Independent vs TAU: Cohen's d: 0.07, -0.41 to 0.55 95% CI, P = .777. GAS and the cognitive score improved for Group + One-to-One vs TAU favoring CR (GAS: Cohen's d: 0.57, 0.19-0.96 95% CI, P = .003; Cognitive score: Cohens d: 0.28, 0.07-0.48 95% CI, P = .008). The QALY costs were £4306 for Group vs TAU and £3170 for One-to-One vs TAU. Adverse events did not differ between treatment methods and no serious adverse events were related to treatment. Both active therapist methods provided cost-effective treatment benefiting functional recovery in early psychosis and should be adopted within services. Some individuals benefited more than others so needs further investigation. ISRCTN14678860 https://doi.org/10.1186/ISRCTN14678860Now closed.